Role of Machine Learning*

Recent advancement of artificial intelligence (AI) and machine learning (ML) at an unprecedented pace is opening a new paradigm for our society. For advanced microelectronics fabrication as well, we expect a data-driven approach will revolutionalize its technological development. In this presentation, I limit our discussion to the application of data-driven science to plasma processing for microelectronics fabrication.[1] AI/ML technologies are likely to have a strong impact especially on three areas of process science and technologies; (1) process development, (2) process control, and (3) fundamental science relevant to plasma processing. For process development, the establishment of databases of surface and gas-phase chemical reactions specific to plasma and thermal processes is crucial. For process control, the formation of “surrogate models” of high-performance computation of plasma models (i.e., digital twins) may allow real-time “virtual metrology (VM),” i.e., plasma diagnostics with a limited amount of system information, to facilitate the existing manufacturing equipment control systems. For fundamental science, opportunities are also limitless. In the presentation, I will summarize the current status of such research activities.     

[1] R. Anirudh, et al., “2022 review of data-driven plasma science” IEEE Trans. Plasma Sci. (2023), in press.